|
Abstract:
|
One of the most common applications of areal data models is the study of disease risk mapping. While several existing methods are very flexible in describing different forms of spatial heterogeneity and correlation, they often do not capture some of the dynamic features, especially when a dataset spans over a long time range. Motivated by the analysis of dengue virus infections data in Puerto Rico, a new model is formulated, with the inclusion of DLM components in addition to a standard CAR prior structure for the spatial random effect. The specification can account for trends, seasonality, and the temporal change in the effect of land coverage, meteorological, and environmental covariates. Estimation is carried using MCMC methods.
|
